Inferred freq bug

[changhiskhan]

In [13]: tsWithGaps.index[3]
Out[13]: <Timestamp: 2001-01-06 00:00:00>

In [14]: tsWithGaps.index[3].dayofweek
Out[14]: 5

In [15]: tsWithGaps.index.inferred_freq
Out[15]: 'B'

[lesteve]

Not sure what happened to the return value of tsWithGaps.index[3] in your snippet but I get this:

In [15]: tsWithGaps.index[3]
Out[15]: <Timestamp: 2001-01-06 00:00:00>

which is what you expect. Not sure how the freq='B' comes into play since I never mentioned anything about a business day frequency but I am guessing that's the bug.

[changhiskhan]

In order to plot with the nice scikits.timeseries style tick labeling, the index is converted from Timestamp to Periods prior to calling Matplotlib

On Jul 24, 2012, at 11:15 AM, lesteve wrote:

Not sure what happened to the return value of tsWithGaps.index[3] in your snippet but I get this:

In [15]: tsWithGaps.index[3]
Out[15]: <Timestamp: 2001-01-06 00:00:00>

which is what you expect. Not sure how the freq='B' comes into play since I never mentioned anything about a business day frequency but I am guessing that's the bug.

---

Reply to this email directly or view it on GitHub:
#1669 (comment)

[changhiskhan]

Sorry, I'm out of it today. This is a false alarm. The issue is not Period constructor. It's in the inferred frequency.
I'll put in a fix

[changhiskhan]

Too much of a performance hit to check in infer_freq
Just check when plotting